Variables
Replace each placeholder before you run the prompt.
{{our_product}}What you sell, the persona, key differentiators.Industrial laser markers built for line-rate traceability. Persona: manufacturing engineer at $50M+ industrial OEM. Differentiators: 24/7 service contract, US-built, integrated I/O.
{{competitor_ads}}Three competitor ads, fully — headlines and descriptions verbatim.Competitor A: "Best laser markers — Save up to 30%! Free shipping. 30-day trial." Competitor B: …
Prompt
You are writing counter-messaging RSAs for a B2B account. Your job is to position against the three competitors below without naming them.
OUR PRODUCT:
{{our_product}}
COMPETITORS:
{{competitor_ads}}
For each competitor, identify their pitch (the angle they're claiming). Then write counter-messaging that:
- Concedes nothing
- Stays in B2B vocabulary (no "save more", no "limited time", no consumer hooks)
- Highlights a specific structural strength of our product that contradicts their angle
- Does NOT name them or quote them
Output:
- 5 distinct headline angles (≤30 chars each)
- Generate 3 headline variants per angle (15 total)
- 4 description variants (≤90 chars each)
- For each headline and description, note the implied competitor angle being countered
Then write a one-paragraph creative brief for the account team explaining the positioning logic.
Do NOT use words like "leading", "premier", "innovative" — they trigger nothing in B2B procurement. Write in the language of the buyer, not the brand.Expected output shape
15 headlines + 4 descriptions, each tagged with the competitor angle countered, plus a brief.
Why we wrote it
B2B copy in agencies usually gets generated in a void — "write me 15 headlines for a laser marker". Anchoring to actual competitor positioning forces the differentiation to be specific.
How to use
- Open Claude or ChatGPT. The recommended model for this prompt is
claude-sonnet-4-6— opus when the prompt requires deep reasoning, sonnet for the rest. - Replace every
{{variable}}with content specific to your account. The examples above are starting points, not templates to ship as-is. - Paste the prompt and run.
- Read the output against the expected shape above. If the model produced a structurally different response, re-prompt rather than accept the drift.